CNN_research

Summer research at Pázmány Péter Catholic University in Budapest, Hungary

Collaborators

My research was funded by a NSF IRES grant and sponsored by Professor Michael Niemier at the University of Notre Dame, who I was in frequent communication with throughout the summer (he was back in the United States). Sophie Lancaster (@slancas1), also an undergrad majoring in Computer Science at Notre Dame, was my research partner. My supervisor at Pázmány in Budapest was András Horvath. András was extremely generous with his time, helping us learn TensorFlow, debug and write code, and understand difficult concepts. We could not have done this research without him.

Our Toolset

We used TensorFlow to build and process neural networks. TensorFlow is a python library that integrates very nicely with NumPy. See here if you are interested in learning to use TensorFlow.

Convolutional Neural Networks

All of the networks we created throughout the research experience were Convolutional Neural Networks (CNNs). CNNs are a type of deep neural network extremely effective at processing, classifying, and recognizing images. This blog provides a very good overview of CNNs.

Our Datasets

• Omniglot: This data can be found in Omniglot_data.

• MNIST: This data can be found in MNIST_data.

• CIFAR-10: This data was too large to upload to GitHub.